🪢 Open source agent evals & observability: Trace, evaluate, and improve LLM applications with one open platform.
TypeScript analyticsautogenevaluationlangchainlarge-language-modelsllama-indexllmllm-evaluationllm-observabilityllmopsmonitoringobservabilityopen-sourceopenaiplaygroundprompt-engineeringprompt-managementself-hostedycombinator
import { describe, expect, it, vi } from "vitest";
import { prepareEvaluatorMetadataForSave } from "./prepareEvaluatorMetadataForSave";
describe("prepareEvaluatorMetadataForSave", () => {
it("generates both missing fields concurrently", async () => {
const generateName = vi.fn().mockResolvedValue(" Quality judge ");
const generateDescription = vi
.fn()
.mockResolvedValue(" Scores response quality. ");
const setName = vi.fn();
const setDescription = vi.fn();
await expect(
prepareEvaluatorMetadataForSave({
currentName: "",
currentDescription: "",
generateName,
generateDescription,
setName,
setDescription,
}),
).resolves.toEqual({
name: "Quality judge",
description: "Scores response quality.",
});
expect(generateName).toHaveBeenCalledOnce();
expect(generateDescription).toHaveBeenCalledOnce();
expect(setName).toHaveBeenCalledWith("Quality judge");
expect(setDescription).toHaveBeenCalledWith("Scores response quality.");
});
it("continues with a generated name when no description is returned", async () => {
const generateName = vi.fn().mockResolvedValue("Quality judge");
const generateDescription = vi.fn().mockResolvedValue(null);
const setName = vi.fn();
const setDescription = vi.fn();
await expect(
prepareEvaluatorMetadataForSave({
currentName: "",
currentDescription: "",
generateName,
generateDescription,
setName,
setDescription,
}),
).resolves.toEqual({
name: "Quality judge",
description: null,
});
expect(generateName).toHaveBeenCalledOnce();
expect(generateDescription).toHaveBeenCalledOnce();
expect(setName).toHaveBeenCalledWith("Quality judge");
expect(setDescription).not.toHaveBeenCalled();
});
it("requires a name when generation returns nothing", async () => {
const generateName = vi.fn().mockResolvedValue(null);
const setName = vi.fn();
const setDescription = vi.fn();
await expect(
prepareEvaluatorMetadataForSave({
currentName: "",
currentDescription: "",
generateName,
generateDescription: null,
setName,
setDescription,
}),
).resolves.toBeNull();
expect(generateName).toHaveBeenCalledOnce();
expect(setName).not.toHaveBeenCalled();
expect(setDescription).not.toHaveBeenCalled();
});
it("fills only missing metadata without overwriting existing text", async () => {
const generateName = vi.fn().mockResolvedValue("Generated name");
const generateDescription = vi
.fn()
.mockResolvedValue("Generated description.");
const setName = vi.fn();
const setDescription = vi.fn();
await expect(
prepareEvaluatorMetadataForSave({
currentName: "Existing name",
currentDescription: "",
generateName,
generateDescription,
setName,
setDescription,
}),
).resolves.toEqual({
name: "Existing name",
description: "Generated description.",
});
expect(generateName).not.toHaveBeenCalled();
expect(generateDescription).toHaveBeenCalledOnce();
expect(setName).not.toHaveBeenCalled();
expect(setDescription).toHaveBeenCalledWith("Generated description.");
});
});